μ‑FTIRReflectance Spectroscopy Coupledwith Multivariate Analysis: A Rapid and Robust Method for Identifyingthe Extent of Photodegradation on Microplastics

Understanding the origins of microplastics (MPs) and evaluating the consequences of plastic pollution require precise chemical information. Moreover, MPs undergo chemical changes due to photoaging, which are worth investigating since they can influence the effects of MPs on living beings and the environment. Micro-Fourier-transform infrared (μ-FTIR) spectroscopy is a key technique for screening MPs, combining optical imaging with chemical information from IR spectra. While reflectance μ-FTIR spectroscopy’s sensitivity to particle thickness and photodegradation complicates automated spectral matching, it can provide valuable information if coupled with multivariate analysis of the data. This study developed a robust method for identifying MPs, even when they are modified by photodegradation. Various acquisition methods (ATR-IR and μ-transflectance-IR), data pretreatments, and data set analysis procedures were examined, and critical aspects were addressed. The proposed method, using μ-TR-IR and principal component analysis (PCA), proved effective for classifying MPs and analyzing their degradation, offering increased sensitivity and a faster workflow compared with manual spectral comparison. μ-TR-IR showed earlier changes in relevant bands, indicating higher sensitivity to degradation than ATR-IR spectroscopy. Despite the notorious issue of spectral artifacts, our results suggest that valuable information can be collected without using sophisticated preprocessing techniques. On the contrary, the presence of the artifacts allows extracting some information on the particles’ thickness. Finally, PCA results were successfully validated for the polymer classification reliability by a test set and compared with the carboxyl index (CI) method to validate the ability to assess degradation. While CI is the most diffused parameter to assess polymer degradation, PCA, which considers the entire spectrum and does not rely on manual integration of single peaks, is inherently more robust than CI and can take into account multiple degradation mechanisms.